Salvato in:
Dettagli Bibliografici
Autori principali: Leandre, Simpenzwe Honore, Shiferaw, Natenaile Asmamaw, Rout, Dillip
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:https://arxiv.org/abs/2508.18960
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866912555205656576
author Leandre, Simpenzwe Honore
Shiferaw, Natenaile Asmamaw
Rout, Dillip
author_facet Leandre, Simpenzwe Honore
Shiferaw, Natenaile Asmamaw
Rout, Dillip
contents In this paper, we propose a vision model that adopts token mixing, sequence-pooling, and convolutional tokenizers to achieve state-of-the-art performance and efficient inference in fixed context-length tasks. In the CIFAR100 benchmark, our model significantly improves the baseline of the top 1% and top 5% validation accuracy from 36.50% to 46.29% and 66.33% to 76.31%, while being more efficient than the Scaled Dot Product Attention (SDPA) transformers when the context length is less than the embedding dimension and only 60% the size. In addition, the architecture demonstrates high training stability and does not rely on techniques such as data augmentation like mixup, positional embeddings, or learning rate scheduling. We make our code available on Github.
format Preprint
id arxiv_https___arxiv_org_abs_2508_18960
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Enhancing compact convolutional transformers with super attention
Leandre, Simpenzwe Honore
Shiferaw, Natenaile Asmamaw
Rout, Dillip
Computer Vision and Pattern Recognition
Machine Learning
In this paper, we propose a vision model that adopts token mixing, sequence-pooling, and convolutional tokenizers to achieve state-of-the-art performance and efficient inference in fixed context-length tasks. In the CIFAR100 benchmark, our model significantly improves the baseline of the top 1% and top 5% validation accuracy from 36.50% to 46.29% and 66.33% to 76.31%, while being more efficient than the Scaled Dot Product Attention (SDPA) transformers when the context length is less than the embedding dimension and only 60% the size. In addition, the architecture demonstrates high training stability and does not rely on techniques such as data augmentation like mixup, positional embeddings, or learning rate scheduling. We make our code available on Github.
title Enhancing compact convolutional transformers with super attention
topic Computer Vision and Pattern Recognition
Machine Learning
url https://arxiv.org/abs/2508.18960